slurm_keys: list
default_opts: dict
sbatch_opt_template: str
sbatch_template: str
sbatch_enter_command: str
sbatch_exit_command: str

__init__(self, description=None, job_name=None, account=None, partition=None, mem=None, nodes=None, ntasks_per_node=None, chdir=None, output=None, steps=(), precall=None, environment=None, **opts): 

LLM Docstring

Build a SLURM batch job from the common #SBATCH options plus any additional ones, the job steps, and optional environment/precall hooks.

• description: str | None

  a human-readable description echoed into the script

• job_name: str | None

  the SLURM job name

• account: str | None

  the SLURM account

• partition: str | None

  the SLURM partition

• mem: Any

  the memory request

• nodes: Any

  the node count

• ntasks_per_node: Any

  tasks per node

• chdir: str | None

  the working directory

• output: str | None

  the output-file pattern

• steps: tuple | str

  the job steps (shell commands)

• precall: Callable | None

  a callable run before writing the file

• environment: dict | None

  environment variables to export

• opts: Any

  additional #SBATCH options

clean_opts(self, opts): 

Makes sure opt names are clean. Does no validation of the values sent in.

• opts: Any

• :returns: _

format_opt_block(self): 

Formats block of options

• :returns: _

format(self): 

Formats an SBATCH file from the held options

• call_steps: Any

• :returns: _

write(self, file, output_dir=None, mode='w+', **kwargs): 

LLM Docstring

Write the formatted SLURM script to a file, optionally within a working directory (into which any precall hook is run).

• file: str | IO

  an open stream or a file path

• output_dir: str | None

  directory to write into (split from file if omitted)

• mode: str

  the file mode when a path is given

• kwargs: Any

  extra arguments for open

• :returns: str | IO

  the path (or stream) written

run(self, file=None, output_dir=None, sbatch_function='sbatch', delete=True, text=True, capture_output=True, *args, **kwargs): 

LLM Docstring

Write the SLURM script and submit it with sbatch (via subprocess.run), optionally deleting the script afterward.

• file: str | None

  the script file name (a temporary name is generated if omitted)

• output_dir: str | None

  directory to write the script into

• sbatch_function: str

  the submission command

• delete: bool

  remove the script file after submission

• text: bool

  run the subprocess in text mode

• capture_output: bool

  capture the subprocess output

• args: Any

  extra positional arguments passed to sbatch

• kwargs: Any

  extra flags passed to sbatch

• :returns: subprocess.CompletedProcess

  the completed-process result